Thoughts on Github copilot

Github copilot is a now popular AI tool that helps developers develop better code by providing suggestions on how to complete your code block

Copilot is powered by Codex, a deep neural network language model trained on public GitHub code repositories.

Most of the coding time is spent designing, debugging, and maintaining the code, not writing the code. Once the code is auto-generated, it’s easy to have more of it. This is not necessarily a problem if all you need to maintain or debug is to change the source from which the code is automatically generated.

It can be a blessing for some and a curse for others. For those who are curses, they may not know for years as they learn less, learn more slowly, increase technical debt, and introduce subtle bugs for years.

Keep in mind that Copilot is an early preview of a very new technology and continues to improve. To see real improvements in program synthesis, go beyond just language models and find more comprehensive solutions that include best practices in human-computer interactions, software development, testing, and many other areas.

Cursed people may not know for years because they learn less, learn more slowly, increase technical debt, and introduce subtle bugs over the years. An experienced programmer working in an unfamiliar language to help you understand typical idioms.

Today, Copilot is more of a product designed and implemented by machine learning researchers than a complete solution with all the necessary expertise.

You can join Github copilot waitlist from here
https://github.com/features/copilot/signup